First thing I want to say: I have been thinking about this and reading what everyone wrote, and I have realized that band coupling can have negative effects (one loud band pulls all others down) if there's too much of it, but still, pulling a single band down that sticks out too much is a good thing. If there's a single band that's really loud it will - if the current coupling is used - not go down as far as it would otherwise have. Which means less consistency. Which is bad. So, maybe the current band coupling should be adjusted to only allow it to pull bands down, but never up. Actually that would make it closer to what the new Band Linking does.
On another note, I have heard this new preset where someone turned Band Coupling off for bands 1 and 2, so a loud sound on band 3 doesn't pull bands 1 and 2 down, which leads to a much warmer sound with more constant bass. I'm now starting to think that what's happening here might be something else: Currently, if ANY band that's coupled to another band has attenuation, all the bands that it's coupled to will be attenuated as well. But, maybe it should only be able to pull a band down if it gets pulled down by BOTH sides (above and below). That would mean that if the audio level falls rapidly above or below a certain frequency, the bands there can ALL go up as much as they need to fill up the spectrum, only if there's a small hole in between the band there would be pulled down by coupling... The effect of this new preset would automatically be achieved by this.
Example to clarify:
Say we have +12 dB audio in bands 1-3 and 0 dB in 4-6
1. Current behavior with coupling: -12 -10 -8 -4 -2 0 - wow, bad!!
2. Behavior with only pulling up: -12 -12 -12 -4 -2 0
3. Behavior where we only respond if BOTH sides pull a band down: -12 -12 -12 0 0 0
In case 1. you get a rise in level close to the frequency where the audio stops. You would get that anyway because band 4 will also contain sound from this band. But it makes the effect bigger.
In case 3. the spectrum is most nicely filled, but I'm not sure which one of 2 and 3 would sound most natural. There is a quite steep bump between band 3 and 4, which might be too much.
Quote:
With coupling method 1 (=adding the 2 control values* and putting the sum to target band) you can accomblish:
-Exactly this, what you described as the reason for putting back the matrix, cause the matrix is using this method with norm = off
So this method (with the table containing the columns i described) could substitute the matrix for easier and more effective usage
Your method 1 is indeed the same as what I have now without Norm(alize). All Norm does it making sure that the sum of all the band couplings for a band is 100%. So if you specify 100% 100% 0% 0% 0%, Norm will internally convert that to 50% 50% 0% 0% 0%. With >100% coupling I find it difficult to properly control the behavior. Combined with what I wrote above (never allow a band to go higher than what you actually calculated for that band) it might get easier.
Quote:
Wit coupling method 2, - the far mor important for me - (compare value 1 and 2 and feed immediately the *higher* one to the target band) you can accomblish:
-Preventing excessive boosts in bands containing no or very less energy and tends to "run away"
-Preventing some "morphing" on critical program content (hearing the re- equalisation working)
-to get sound results nearer to the original and therefore add fidelity to the sound
-and a little bit to turn away the increase in multiplex power in the case of a substantially growth of energy in the middle band(s)
I don't understand what you mean by the increase in multiplex power?
To stay close to the original, the current coupling might actually work pretty well (maybe too well). As I described above.
Quote:
Method 2 becomes more significant with a) higer MB Drives and b) steeper crossovers
Quote:
The goal in my opinion is not primary to prevent a single band to stick out (even this is always partially a side effect of coupling), it *should* stick out if a loud sound comes in this area to compress such sound. And here is the reason for doing also method 2 in percent: If a single band gets a lot of energy and is therefore perforimg a high gain reduction (particular higher than in normal cases), the resulting *absolute* coupling is less - the bands coupled to the sticking out band are not pulled so deep into gain reduction (wich would be more audible, cause the output becomes quieter instead of louder).
I'm doing the opposite now though - coupling (not linking!) works in dB's. If you have a band that sticks out a lot (much less attenuation), working in dB's will pull it down further which is good. On the other hand, if a single band goes down a lot, using % would be better because it would less strongly affect the surrounding bands.
Quote:
Doing method 2 in dB would make sense in the AGC: So you can determine that the bass band is allowed to boost the bass by e.g. 6dB compared to the rest of the spectrum - and this *independent* from the input level and therefore from the Gain reduction.
Skipping this part for now... I first need to fully understand what to do with MB.
Quote:
Furthermore is the point i described relating to the attack and release timing affecting the coupling:
When you see method 1 (and also the coupling matrix- method) and having the following case:
You have put an amount of percent of band 3´s control value to band 2. And you have set band 2 to a slower release time than band 3. Now a loud sound causes band 3 to go deep into gain reduction. Cause of the coupling band 2 follows band 3. Now the sound energy in band 3 disappears suddenly. Band 2 should go as fast as band 3 (and not with band 2´s release time) back to the lower amount of gain reduction.
This is already the case in both Band Coupling and Band Linking.
Quote:
And yes, you are right - the primary direction used in coupling is *from* the middle band(s) *to* the bass bands and *to* the high end bands. I would never couple band 1 to 2, or band 5 to four, or band 4 to 3 in a 5-band preset. But also all that should be possible (for e.g. artifactional sounds) - and it would be with the described columns/rows method.
If you want to protect a band from sticking out you need to look in both directions, always.
Quote:
The last point is maintaining the adding- method, how the values are added.
In the matrix with norm = off the control values are added directly - right? Therefore you get such huge gain reductions in the target band unless you are using really less coupling values. It should be added by vector, so when you see the both values as orthographical vectors than the length of the third vector (the bisecting line) shows the result.
This is indeed how they are added - I don't understand what you mean by those vectors though because it's just a single value, not a complex (x,y) value. So why a vector?
Register
Login
FAQ
Search
